Journal list menu

Volume 593, Issue 15 p. 1944-1956
Research Article
Free Access

Biochemical study of sortase E2 from Streptomyces mobaraensis and determination of transglutaminase cross-linking sites

Anita Anderl

Anita Anderl

Department of Chemical Engineering and Biotechnology, University of Applied Sciences of Darmstadt, Germany

Department of Chemistry, Technische Universität Darmstadt, Germany

Search for more papers by this author
Cathrin Ferlemann

Cathrin Ferlemann

Department of Chemical Engineering and Biotechnology, University of Applied Sciences of Darmstadt, Germany

Search for more papers by this author
Marius Muth

Marius Muth

Department of Chemical Engineering and Biotechnology, University of Applied Sciences of Darmstadt, Germany

Bioengineering and Biosystems, Institute of Functional Interfaces, Karlsruhe Institute of Technology, Germany

Search for more papers by this author
Antonina Henkel-Gupalo

Antonina Henkel-Gupalo

Department of Chemical Engineering and Biotechnology, University of Applied Sciences of Darmstadt, Germany

Search for more papers by this author
Aileen Ebenig

Aileen Ebenig

Department of Chemistry, Technische Universität Darmstadt, Germany

Search for more papers by this author
Gerald Brenner-Weiß

Gerald Brenner-Weiß

Bioengineering and Biosystems, Institute of Functional Interfaces, Karlsruhe Institute of Technology, Germany

Search for more papers by this author
Harald Kolmar

Harald Kolmar

Department of Chemistry, Technische Universität Darmstadt, Germany

Search for more papers by this author
Hans-Lothar Fuchsbauer

Corresponding Author

Hans-Lothar Fuchsbauer

Department of Chemical Engineering and Biotechnology, University of Applied Sciences of Darmstadt, Germany

Correspondence

H.-L. Fuchsbauer, Department of Chemical Engineering and Biotechnology, University of Applied Sciences of Darmstadt, Stephanstraße 7, Darmstadt 64295, Germany

Tel: +49 6151 1638181

E-mail: [email protected]

Search for more papers by this author
First published: 03 June 2019
Citations: 2
Edited by Peter Brzezinski

Abstract

Distinct streptomycetes such as Streptomyces mobaraensis produce the protein cross-linking enzyme transglutaminase. Bioinformatic analysis predicted the occurrence of seven sortases exerting transpeptidation reactions similarly to transglutaminase. Here, we report the production and characterization of sortase E2 (Sm-SrtE2) solubilized by removal of its membrane anchor domain. Sm-SrtE2 activity was measured using pentapeptides predicted to be cell wall sorting signals of putative sortase substrate proteins. Preferred linkage to Gly3 by Sm-SrtE2 was in the order LAETG>>LAHTG>>LAQTG~LANTG>LARTG. Chaplin 1 from S. mobaraensis was further demonstrated to be an excellent substrate of both the intrinsic Sm-SrtE2 and transglutaminase. The unexpected discovery showing Gln-62 and Gln-65 of Δ1–50-Sm-SrtE2 as transglutaminase cross-linking sites suggests that low enzyme stability might be due to anchor domain truncation and a disordered N terminus.

Abbreviations

abz, 2-aminobenzoyl

Chp, chaplin

CWS, cell wall sorting signal

dab (dabcyl), 4-dimethylaminoazobenzene-4′-carboxylic acid

dnp, 2,4-dinitrophenyl

DAIP, dispase autolysis-inducing protein

Edans (ans), 5-(2-aminoethylamino)naphthalenesulfonic acid

IDA, independent data acquisition

IMAC, immobilized metal-ion affinity chromatography

MBC, monobiotin cadaverine

MTG, transglutaminase

Rdl, rodlin

Srt, sortase

SUMO, small ubiquitin-related modifier

TAMRA, 5-carboxytetramethylrhodamine

Sortases (Srts, EC 3.4.22.70) are extracellular enzymes that covalently link surface proteins to the peptidoglycan, even contributing to bacterial pili assembly in gram-positive bacteria [1]. Bioinformatic analyses of bacterial genomes revealed the occurrence of six subfamilies, denominated as SrtA–F, which may considerably vary in structure and preference for distinct substrate-binding motifs [2-4]. All sortases appear to be linked to the outside of cytoplasmic membranes via N-terminal anchor peptides [1]. The residual protein then adopts the canonical eight-stranded β-barrel fold of sortases, but structural variations, relevant for both terminal peptides and two flexible loops, are likely to modulate substrate specificity [5-9]. The active site, shaped by the catalytic triad His-Cys-Arg, is arranged in close vicinity to the substrate-binding pockets. The sortase substrates typically harbor a five-residue-recognition motif, the so-called cell wall sorting signal (CWS), a hydrophobic amino acid strand, and a positively charged tail that arrests the C-terminal domain in the cytoplasmic membrane. In Staphylococcus aureus and many other gram-positive bacteria, class A sortases (e.g. Sa-SrtA, UniProt Q2FV99, PDB 1IJA) use the LPXTG motif, cleave the peptide bond between threonine and glycine, remove the C-terminal membrane domain, and form a new peptide bond with the cross-bridging peptide of lipid II, the peptidoglycan precursor molecule [10, 11]. Influence of metal ions on catalysis rate has been shown as to be negligible, but Ca2+ considerably enhances transpeptidation rate in Sa-SrtA [5]. Substrates of ‘housekeeping’ sortases such as Sa-SrtA exert various functions, and their genes are dispersed over the whole chromosome [3]. More specialized sortases, commonly encoded by genes co-localized with a substrate gene, mediate the transpeptidase reaction with usually a single substrate protein [2].

The gram-positive, multicellular streptomycetes, dwelling in the soil under decomposing organic matter [12], likewise use sortases for the attachment of surface proteins. The genome of the reference strain Streptomyces coelicolor A3(2) includes several sortase genes, two for class E and five for class F enzymes [3]. The two genes (SCO3849, SCO3850) for Sc-SrtE1 (UniProt Q9XA14, PDB 5CUW) and Sc-SrtE2 (Q9XA15) are juxtaposed in a conserved gene region that occurs in many Streptomyces bacteria [13]. The lack of adjacent genes encoding CWS-containing proteins has led to the assumption that SrtE enzymes are transpeptidases with ‘housekeeping’ function [13]. Contrary to sortases A, the SrtE enzymes prefer the CWS motifs LAETG and LAHTG and cleave, less specifically, the peptide bonds between both Thr-Gly and Ala-Glu/Ala-His [13]. Bioinformatic analyses predicted up to 14 Sc-SrtE and five Sc-SrtF sortase substrates, among them long chaplins (Sc-ChpA, SCO2716, UniProt Q8CJY7; Sc-ChpB, SCO7257, Q9X7U2; Sc-ChpC, SCO1674, Q9AD93) and a cellulose-binding protein (SCO1734, Q9S296) [2-4, 13, 14]. Chaplins (Sc-ChpA–H) and rodlins (Sc-RdlA–C) assemble to amyloid-like fibrils at the aerial hyphal surface of streptomycetes, thus forming the characteristic rodlet pattern [15-17]. Cell wall anchoring of Sc-ChpC and recognition of the LAXTG motif by Sc-SrtE1/E2-mediated cleavage of artificial peptides have been proved in vivo and in vitro [13].

Members of the large and heterogeneous Streptomyces genus differ from S. coelicolor A3(2) in morphology and many functions [18]. For instance, Streptomyces mobaraensis, formerly Streptoverticillium mobaraense, forms verticil-like instead of spiral aerial hyphae and cross-links extracellular proteins by the enzyme transglutaminase (Sm-MTG, EC 2.3.2.13, UniProt P81453, PDB 1IU4), which is absent in S. coelicolor [19, 20]. Moreover, lack of genes for distinct Sm-MTG substrates such as the well-studied dispase autolysis-inducing protein (Sm-DAIP, UniProt P84908, PDB 5FZP) in the genomes of many streptomycetes underlines the special character of S. mobaraensis [21, 22]. Protein cross-linking by Sm-MTG proceeds in a transpeptidation-like reaction but, contrary to sortases, between the side-chains of protein-bound glutamine and lysine residues, resulting in the formation of intra- and intermolecular cross-bridges [23]. The occurrence of two enzymes that contribute to the development of irreversibly cross-linked glycoside–protein or protein–protein networks in the cell wall of S. mobaraensis raises the question, what are their specific functions?

Here, we report the production of SrtE2 from S. mobaraensis in Escherichia coli. SrtE2 is regarded as being more important for protein anchoring than SrtE1 in S. coelicolor [13]. Low protein stability caused us to prepare three variants differing in the N-terminal anchor domain. Enzyme features and transpeptidation rates were examined using intrinsically quenched pentapeptide fluorophores and two long chaplins, Chp1 from S. mobaraensis and ChpC from S. coelicolor. Interestingly, both chaplins and Sm-SrtE2 turned out to be glutamine donor substrates of Sm-MTG. Although Sm-SrtE2 labeling is likely the result of N-terminal protein truncation, the study provides additional insights into MTG-mediated protein conjugation.

Materials and methods

Protein expression and purification

The gene encoding Sm-SrtE2 without N-terminal anchor was codon-optimized by GenScript (Piscataway, NJ, USA) and inserted into pET28a vector. Mutagenesis for exchange of glutamines for asparagines was performed by splicing by overlap extension PCR. E. coli BL21 (DE3) plysS (Merck-Millipore, Darmstadt, Germany) transformed by pET28a-Sm-srtE2 was grown in LB medium at 37 °C to D600 of 0.4–0.6, prior to the induction of protein production with 0.2 mm IPTG at 28 °C. Supernatants of sonicated cells in 50 mm Tris/HCl, 300 mm NaCl, 15 mm imidazole, 10% glycerol pH 8 were separated by 1 mL Ni2+-chelating Sepharose Fast Flow using stepwise and linear increasing imidazole of 15 mm (20 mL), 25 mm (20 mL), 50 mm (15 mL) and 100–500 mm (5 mL). Sm-SrtE2 fractions were pooled and further purified by Fractogel EMD urn:x-wiley:00145793:media:feb213466:feb213466-math-0001 chromatography using 50 mm citrate pH 6 and linear increasing NaCl of 0–1 m.

The chaplin production was carried out in E. coli BL21 (DE3) plysS, transformed by codon-optimized genes for Sm-Chp1 or Sc-ChpC fused to N-terminal His6–small ubiquitin-related modifier (SUMO) in pET14b. In brief, after cell supernatants were incubated with immobilized metal-ion affinity chromatography (IMAC) resin, washing steps were performed as described above, and the fusion proteins were eluted by 50–500 mm imidazole. Upon dialysis and removal of the His6–SUMO by SUMO protease (molar E/S ratio of 1 : 1000) in 50 mm Tris/HCl, 300 mm NaCl pH 8 for 2 h at room temperature, separation by IMAC (flow through) and Superdex 75 size exclusion chromatography in 25 mm HEPES/100 mm NaCl pH 8 resulted in highly purified proteins.

Determination of protein melting points by differential scanning fluorimetry

To analyze protein stability, melting points of 0.2 mg·mL−1 truncated Sm-SrtE2 were determined using a Prometheus NT.48 instrument (Nanotemper Technologies, Munich, Germany). While increasing the temperature by 1 °C·min−1, the ratio of intrinsic fluorescence at 350 and 330 nm was monitored. The first derivative of this function was used to calculate the melting points.

Activity assay of Sm-SrtE2

Enzyme activity was continuously measured in black half area 96-well microplates (Greiner bio-one, Bischofsheim, Germany) using various amounts of internally quenched fluorescent pentapeptides of the type dabLAXTGans (X = E, H, N, Q, R; GenScript) and 1 μm Sm-SrtE2 in 150 μL 50 mm Tris/HCl, 100 mm NaCl, 10% glycerol pH 7.0 at 30 °C. The increase in fluorescence at 520 nm (λexc of 340 nm) was monitored every 30 s using Polarstar multi-mode microplate reader (BMG Labtech, Ortenberg, Germany). For determining the kinetic parameters, initial velocities were estimated from linear slopes using 232 rfu (X = H), 344 rfu (X = N), 421 rfu (X = E), 522 rfu (X = R) or 574 rfu (X = Q) per 1 nmol·mL−1 released Gly-Edans and fitted by prism (GraphPad Software Inc., La Jolla, CA, USA).

Analysis of the CWS pentapeptide cleavage site

An Agilent 1100 HPLC system (Agilent, Waldbronn, Germany) coupled to an X500R quadrupole time-of-flight mass spectrometer with Turbo V ESI source (Sciex, Darmstadt, Germany) was used for peptide analysis. Prior to HPLC separation, Sm-SrtE2 was precipitated with methanol from overnight reaction mixtures containing 50 μm dabLAXTGans (X = E, H, N, Q, R). Twenty microliters of 0.5 μm dabLAXTGans was separated by a linear gradient of acetonitrile from 10% to 90% in 0.1% (v/v) formic acid for 15 min at flow rates of 0.7 mL·min−1 using a 100 × 4 mm Hypersil ODS C18 3 μm 120 Å column (Thermo Fisher, Darmstadt, Germany). MS data were acquired by independent data acquisition (IDA) at m/z of 100–1100 Da.

Sequence analysis of long chaplins

The long chaplins were excised from polyacrylamide gels and, upon destaining (1 : 1 mixture of acetonitrile/10 mm NH4HCO3, 10 min) and reducing (10 mm DTT/10 mm NH4HCO3, 60 °C, 15 min), alkylated by 55 mm iodoacetamide for 15 min at room temperature and digested by 10 ng·μL−1 trypsin in 10 mm NH4HCO3 pH 8.0. The obtained peptide mixture was analyzed by the same HPLC-MS/MS system and eluents as described previously. Peptides were eluted from a TSKgel Super-ODS C18 column 100 × 2 mm, 2.3 μm particle size, 140 Å pore size (Tosoh, Griesheim, Germany) using a linear gradient starting from 10% to 35% acetonitrile in 55 min at flow rates of 0.2 mL·min−1. MS/MS data were acquired by IDA at m/z 300 to 2800 Da and the collected data were analyzed using proteinpilot 5.0.1 (Sciex, Darmstadt, Germany).

Fluorescence labeling of chaplins

Labeling of chaplins was achieved by incubating 10 μm chaplin and 50 μm Gly3-TAMRA with 1 μm Sm-SrtE2 in 50 mm Tris/HCl, 100 mm NaCl, 10% glycerol pH 7 at room temperature The reaction mixtures were heated at various times in SDS application buffer, separated by SDS/PAGE and illuminated at 312 nm.

Protein biotinylation using MTG

Sm-MTG-mediated incorporation of (mono)biotin cadaverine (MBC; Zedira, Darmstadt, Germany) was performed as described [21]. The reaction mixture containing 5 μm Sm-SrtE2, 0.2 mm MBC and 0.1 μm Sm-MTG was incubated in 50 mm Tris/HCl, 100 mm NaCl, 10% glycerol pH 7 at room temperature The molar Sm-MTG Sm-Chp1/Sc-ChpC ratio was 1 : 5000. Reaction mixtures were heated at various times in SDS application buffer, separated by SDS/PAGE, blotted on poly(vinylidene difluoride) membrane and stained by streptavidin-IRDye 800 conjugates (LI-COR Biosciences, Bad Homburg, Germany). Equal areas of fluorescent bands were integrated to evaluate the degree of biotinylation using the Odyssey® SA infrared imaging system (LI-COR Biosciences).

Synthesis of Gly-Edans

Aliquots of 1.65 mmol dicyclohexyl carbodiimide were added to a stirred mixture of 1.5 mmol BOC-Gly and 1.8 mmol pentafluorophenol in 4.5 mL dioxane/dimethylformide (3 : 1) at −3 °C for 30 min. Upon additional stirring for 2 h, the precipitated solid was removed by filtration to obtain the pentafluorophenyl ester of BOC-Gly (BOC-Gly-OPFP).

A solution of 0.5 mmol 5-(2-aminoethylamino)naphthalenesulfonic acid (Edans) in 1.5 mL anhydrous ethanol containing 1 m NaOH was evaporated to dryness, dissolved in 50 mL dimethylformamide, mixed with BOC-Gly-OPFP and stirred overnight. The mixture was reduced to 3 mL in vacuo and separated by SiO2 chromatography (60 cm × 3 cm column). Fractions containing highly purified BOC-Gly-Edans were combined, evaporated and washed several times with diethyl ether.

42.3 mg (0.1 mmol) BOC-Gly-Edans in 1.5 mL CH2Cl2 and 1.8 mL trifluoroacetic acid was stirred at ambient temperature for 1 h. Upon evaporation to dryness, the residue was washed with diethyl ether and dried in vacuo. Structure of Gly-Edans was confirmed by 1H-NMR (300 MHz, DMSO-D6).

Results

Bioinformatic studies

The shot-gun-sequenced genome of S. mobaraensis (266 contigs) allowed the discovery of seven sortase-encoding genes [24]. As was shown in S. coelicolor A3(2) and other streptomycetes [13], the same conserved DNA cluster on contig7 accommodates genes for Sm-SrtE1 (UniProt M3BQ44) and Sm-SrtE2 (M3BQ44/M3CCT9) but not for a third Srt, Sm-SrtE3 (Fig. S1A). For comparison, the transglutaminase gene on contig57 is in close vicinity to highly conserved genes putatively encoding aconitase (M3C562) and UDP-N-glucosamine 1-carboxyvinyltransferase (M3BHP9). Both enzymes but not transglutaminase are encoded by genes (SCO5999, SCO5998) of S. coelicolor (Fig. S1B).

Though great differences between S. coelicolor and S. mobaraensis exist, sequence homology of SrtE1 and SrtE2 was comparably high, sharing 44% and 61% identical amino acids respectively (Fig. S2). Moreover, none of the genes within the Sm-SrtE cluster encodes a CWS-containing protein, thus suggesting the dispersion of Sm-SrtE substrates over the whole S. mobaraensis chromosome as is the case in S. coelicolor [2-4].

Screening for CWS-containing proteins using the sequence motif LAHTG resulted in more than 100 hits, which were further assessed, especially with respect to sequence and putative extracellular function (Table 1). Interestingly, the genome of S. mobaraensis has only a single gene for a ChpC-like long chaplin (Sm-Chp1, UniProt M2ZW51) that is most related to a putative protein from Streptomyces caatingaensis (Table 1, Figs S1C and S3). Many of the predicted S. coelicolor sortase substrates were not available in S. mobaraensis, and homology between the remaining proteins was generally low. While S. coelicolor sortases seem to prefer histidine or glutamate in the middle of the pentapeptide LAXTG, the occurrence of His-, Glu-, Arg-, Gln- or Asn-containing CWS motifs suggests broader specificity of Sm-SrtE2. A noteworthy fact may be that CWS-like pentapeptides are distributed over the whole substrate protein sequences and were even found in N-terminal or central protein regions. For instance, the putative, Q9KYJ0-related (SCO2682 gene) terpenoid cyclase/protein prenyltransferase M3ABC7 harbors two LAXTG motifs, LASTG 49 aa downstream from the predicted 33 aa signal peptide and LARTG 25 aa from the Arg-rich C-terminal peptide (Fig. S3). In the putative pyridoxal phosphate-dependent aminotransferase M3B5F0, LAETG and the hydrophobic stretch are parts of the predicted coenzyme- and substrate-binding domain and, thus, unlikely to be a substrate of sortase E.

Table 1. Putative substrates of the sortases E1 and E2 from Streptomyces mobaraensis.
Sortase substratea Putative function Sorting signal C-terminal peptide sequence Most related proteina S. coe. geneb, proteina
M2ZW51 Long chaplin (Chp1) (aerial hyphae surface) LAHTG GAVLYRRSRAGQQG

S. caatingaensis

A0A0K9X8D7

SCO1674
N1NTQ8 Q9AD93
M3CEA1 Iron adsorption and transport LAHTG AVWTARRKSAKLTG

S. caatingaensis

A0A0K9XHV1

SCO2270
Q9RKQ8
M3B5F0 Pyridoxal phosphate-dependent aminotransferase LAETG IRSRPDRSLRQVSR

S. platensis

A0A1Y2NJ80

n. f.c
M3CCZ8 M6 family, thermolysin-like metalloprotease LANTG RTATAICTLRKSGR

S. albireticuli

A0A1Z2LC49

n. f.c
M3ABC7 Terpenoid cyclase, protein prenyltransferase LARTG AVVVSRRRRRDDAG

S. harbinensis

A0A1I6PD91

SCO2682
Q9KYJ0
M3C0P0 [P(Q/K)PPT(T/E)D]10 repeat-containing protein LAQTG FRLLPRAINRNTVA

S. caatingaensis

A0A0K9X7A3

SCO2971
Q9L1S4
M2BEI1 Short chain dehydrogenase LARTG AGAGVAGGLRRARR

S. caatingaensis

A0A0K9XKV0

SCO7413
Q9L146
M2ZV31 CAAX prenyl protease-like protein (M79 family) LARTG VGARASGPVVPDRL

S. sparsogenes

A0A1R1SN31

n. f.c
  • aUniProt protein identification numbers. bGene numbers of Streptomyces coelicolor A3(2). cn. f., not found in the annotated genome of S. coelicolor A3(2).

Production, stability and functionality of sortase E2

Several E. coli strains were transformed using various plasmids and a codon-optimized gene shortened by DNA for the putative Sm-SrtE2 membrane anchor domain. The expected Δ1–50-Sm-SrtE2 proteins were N- and C-terminally fused to His6/10 tags, SUMO or glutathione-S-transferase to support purification. A pelB leader sequence allowed the protein to transport to the periplasm. Nearly all attempts resulted in large amounts of inclusion bodies while cell supernatants contained at most traces of the soluble enzyme that slowly hydrolyzed the CWS-like peptide abzLAHTGdnp. However, E. coli BL 21 (DE3) or E. coli BL 21 (DE3) pLysS, transformed by pET-28a, produced significant amounts of N-terminally His6-tagged Δ1–50-Sm-SrtE2, depicted by thin protein bands upon adsorption of cell supernatants from 2 L cultures to Ni-iminodiacetic resins (not shown). This protein tended to form precipitates, already at 4 °C, and was lost during purification. We decided to remove completely the predicted N-terminal helix domain to obtain a more soluble Sm-SrtE2 variant (Fig. 1). Using the same expression systems, the yields of Δ1–55- and Δ1–78-Sm-SrtE2 were still lower, and protein stability was not enhanced (Fig. 1). Both proteins were active and cleaved the reporter peptides abzLAHTGdnp or dabLAHTGans. We then succeeded in increasing the yield of Δ1–50-Sm-SrtE2 up to 0.5 mg protein per liter culture and completed purification of the IMAC-concentrated enzyme by Fractogel EMD urn:x-wiley:00145793:media:feb213466:feb213466-math-0002 chromatography (CEX) that removed residual E. coli proteins (Fig. 2A). Further concentration up to 6 mg·mL−1 for crystallization attempts was associated with loss of Δ1–50-Sm-SrtE2 by the formation of insoluble aggregates. Additives such as 0.1 mg·mL−1 BSA or glucose, galactose, saccharose, lactose, trehalose, maltodextrin, Triton X-100, CHAPS, N-lauroyl sarcosine and dodecyl-β-d-maltoside, 0.1% each, had no significant effect on protein stability. Storage at −20 °C and −80 °C were the best method of maintaining Sm-SrtE2 activity. It should be noticed that Sm-SrtE2 forms, like Sa-SrtA [25], a dimer under non-reducing conditions (Fig. 2A, line CEX/nr).

Details are in the caption following the image
Predicted structure of the sortase E2 variants from Streptomyces mobaraensis. The Sm-SrtE2 model was built by phyre2 and chimera 1.13.1rc. The removed N-terminal peptides are depicted in sea blue (Δ1–50-Sm-SrtE2), violet (Δ1–55-Sm-SrtE2), and medium blue (Δ1–78-Sm-SrtE2), the active site residues in red, and the glutamines in orange. The purified proteins, characterized by Coomassie-stained 12.5% SDS/PAGE and melting points, are shown on the right (n. d., not determined).
Details are in the caption following the image
Protein patterns of purified sortase E2 (Δ1–50-His6-Sm-SrtE2), Sm-chaplin 1 and Sc-chaplin C. (A) Sm-SrtE2 was purified by Ni-IDA (IMAC) and Fractogel EMD urn:x-wiley:00145793:media:feb213466:feb213466-math-0003 (CEX) chromatography. Lane CEX/nr, CEX fraction under non-reducing conditions. (B, C) Purification of Sm-Chp1 (B) or Sc-ChpC (C) as His6-SUMO fusion proteins by (a) IMAC, SUMO protease treatment (PT), (b) IMAC (SUMO removal), (c) IMAC (only Sm-Chp1 via its His-rich regions), and SEC. The combined fractions were separated by 12.5% SDS/PAGE and Coomassie-stained.

We further produced chaplin 1 from S. mobaraensis (Sm-Chp1) and chaplin C from S. coelicolor A3(2) (Sc-ChpC) in E. coli BL21 (DE3) pLysS as SUMO fusion proteins to study protein interaction with Sm-SrtE2 and microbial transglutaminase (Sm-MTG). SDS/PAGE usually displayed a single but broad Sm-Chp1 and two distinct Sc-ChpC bands near the 40 kDa marker protein (Figs 2B,C and 4C). LC-MS/MS analysis proved that the chaplin bands were caused by the same protein and were most likely the result of various conformations (Fig. S4). A comprehensive report on both chaplins is forthcoming.

Kinetics of sortase E2

The bioinformatic analysis has shown that at least five different LAXTG (X = E, H, N, Q, R) cell wall sorting motifs (CWS) may exist in the sequence of putative sortase substrates from S. mobaraensis (Table 1). In order to study Sm-SrtE2 activity and specificity, we used the corresponding pentapeptides linked to fluorescent dyes (2-aminobenzoic acid, abz, or 5-(2-aminoethylamino)naphthalenesulfonic acid, ans) and quenching molecules (2,4-dinitrophenol, dnp, or 4-dimethylaminoazobenzene-4′-carboxylic acid, dab) following the previously described Sa-SrtA activity assays [26, 27]. All pentapeptides (dabLAXTGans) turned out to be substrates of Δ1–50-Sm-SrtE2 and allowed us to examine the preferred cleavage site by LC-MS. Although Sm-SrtE2-mediated dabLAETGans hydrolysis was performed by incubation overnight, only two product peaks were found, which were clearly caused by the tetrapeptide dabLAET and Gly-EDANS (Fig. 3A, Fig. S5). The sole cleavage site between threonine and glycine proved that Sm-SrtE2 is more similar to Sa-SrtA than to Sc-SrtE2 in this respect. Sc-SrtE2 from S. coelicolor has been reported to either cleave the Ala-Xaa (Xaa = Glu, His) or the Thr-Gly peptide bond within the CWS peptide [13].

Details are in the caption following the image
Activity and specificity of sortase E2 (Δ1–50-Sm-SrtE2) from Streptomyces mobaraensis. The assay was performed by incubating 1 μm Sm-SrtE2, 40 μm dabLAXTGans and 5 mm Gly3 in 50 mm Tris/HCl, 100 mm NaCl, 10% glycerol, pH 7 at 30 °C for 10 min. (A) HPLC of dabLAETGans (blue) and after overnight incubation with Sm-SrtE2 (red). The molecular masses of substrate and hydrolysis products are depicted by m/z ratios. (B) Relative activity of Sm-SrtE2 using the indicated peptides. The increase in fluorescence of 457 rfu·min−1 was defined as 100% activity. (C) Non-linear plot for estimating Sm-SrtE2 enzyme parameters using dabLAETGans. Inset: increase in fluorescence at the used pentapeptide concentrations. (D) Non-linear plot using dabLAETGans and Gly-NH2 in the indicated concentrations. Inset: relative activity of Sm-SrtE2 in the absence (−) or presence of Gly-NH2 and Gly3. Activity obtained in the presence of Gly3 was set to 100%. (E) Influence of pH on Sm-SrtE2 activity using acetate pH 3.5 (purple triangle), citrate pH 4.5–7.5 (blue circles), phosphate pH 6.0–8.0 (green triangles) or Tris/HCl pH 7.0–9.0 (rusty-red rhombi). Activity obtained at pH 7 was set to 100%. (F) Inhibition of Sm-SrtE2 by Zn2+ in the indicated concentrations. Inset: influence of the indicated metals (5 mm) on Sm-SrtE2 activity. Activity obtained in the absence of metals was set to 100%. All data were performed in triplicates and fitted by prism. The standard deviations are shown by vertical bars.

Sm-SrtE2 was then used to investigate fragment quenching effects on transpeptidation reactions using the dab/ans pentapeptides and triglycine. To this end, mixtures, containing various amounts of each pentapeptide, were incubated overnight, thus completing the release of fluorescent Gly-Edans. We observed approximately linear increase in fluorescence but only at low substrate concentrations (Fig. S6A). While triglycine substitution of up to 40 μm LARTG, LAHTG and LANTG was roughly correlated with the fluorescence intensity, quenching of the released Gly-Edans occurred at still lower concentrations by dabLAET or dabLAQT. Moreover, increase in fluorescence depended on the central amino acid ranging from 232 rfu·μm−1 (LAHTG) to 574 rfu·μm−1 (LAQTG) cleaved pentapeptide (Table 2, Fig. S6). For comparison, fluorescence of Gly-Edans was 490 rfu·μm−1.

Table 2. Kinetic parameters of sortase E2 (Δ1–50-Sm-SrtE2) using dabLAXTGans. Increase in fluorescence was continuously monitored by incubating 1 μm SrtE2, 10–320 μm dabLAXTGans and 5 mm Gly3 in 50 mm Tris-HCl, 100 mm NaCl, 10% glycerol, pH 7.0 at 30 °C and 520 nm (λExc of 340 nm). The relative fluorescence units (rfu) per μm cleaved pentapeptide (Fig. S6) were used to calculated kcat. Fluorescence of Gly-Edans was 490 rfu·μm1
dabLAXTGans (X) Fluorescence (rfu·μm−1) Kmm) kcat (h−1) kcat/Km (mm−1·h−1)
E 421 49.4 ± 10.6 92.4 ± 5.9 1870 ± 521
H 232 46.5 ± 12.1 56.2 ± 8.7 1208 ± 419
Q 574 37.5 ± 8.2 24.6 ± 6.7 654 ± 186
N 344 172 ± 62 94.0 ± 15.2 546 ± 293
R 522 71.6 ± 13.3 16.5 ± 6.3 231 ± 58

Using this assay, preference of Sm-SrtE2 was in the order LAETG>>LAHTG>>LAQTG~LANTG>LARTG (Fig. 3B,C). As has been shown for Sa-SrtA [26, 27], affinity of Sm-SrtE2 for the pentapeptides was high but the turnover rate low, even if Gly3 was present in the assay (Table 2, Fig. 3C). Without acyl acceptor, catalytic efficiency of Sm-SrtE2 was still lower, and nucleophilic compounds such as Gly-NH2, Gly-OEt, Gly3 or H2NOH considerably accelerated the catalysis. The most effective molecule was Gly-NH2, achieving maximum Sm-SrtE2 transpeptidation rate at concentrations of about 100 μm (Fig. 3D). It should be noted that dabLAXTGans pentapeptide cleavage by Sm-SrtE2 was only moderately correlated with the linearity of fluorescence increase, thus limiting the comparison with other sortases.

Sortase E2 from S. mobaraensis has a broad pH optimum ranging from pH 5.5 to pH 7.5 (Fig. 3E). In contrast to sortase A, calcium ions had no or a slightly depressive effect on Sm-SrtE2 activity (Fig. 3F, inset). Heavy metals such as zinc, copper, cobalt or nickel inactivated the enzyme, most likely by their affinity to the catalytic cysteine. At a concentration of 11.5 μm, zinc reduced enzyme activity by half (Fig. 3F).

Specificity of sortase E2 for chaplins

The long chaplin from S. mobaraensis, Sm-Chp1, accommodates the CWS motif LAHTG, thus appearing, in the light of the pentapeptide transpeptidation results, as a moderate substrate of Sm-SrtE2. The LAHTG sequence is conserved among putative Chp1-like proteins from transglutaminase-producing streptomycetes (Fig. S7). By comparison, S. coelicolor has evolved three chaplins, Sc-ChpA, Sc-ChpB and Sc-ChpC, that may exert interaction with sortase via LAHTG and LAETG binding sites. We used Sc-ChpC, most related to Sm-Chp1 but dissimilar in the CWS motif, to further examine significance of the central amino acid of the LAXTG peptides (Fig. 4A, Fig. S8). Progress in the transpeptidation reaction was tracked by the Sm-SrtE2-mediated incorporation of Gly-Edans or Gly3-TAMRA that allowed visualization by irradiation upon electrophoresis (Fig. 4B,C). While Gly-Edans caused scarcely visible protein bands (not shown), linkage to the more sensitive TAMRA peptide resulted in strongly labelled Sm-Chp1 (Fig. 4C, left panel). The broad band, most likely caused by conformationally disparate isomers of Sm-Chp1, was consistently stained by the Sm-SrtE2-mediated fusion with Gly3-TAMRA. Incubation for 5 min already yielded a strong fluorescent band, and conjugation was nearly completed after 60 min. In contrast, while showing two distinct protein bands, only the slower migrating Sc-ChpC protein was linked to the fluorescent peptide, first visible after 1 h (Fig. 4C, right panel). It is important to remember that LC-MS/MS clearly showed occurrence of the CWS motifs in both Sc-ChpC isomers (Fig. S4). As removal of the LAETG peptide must be excluded, failure of Sm-SrtE2-mediated labeling of the faster migrating Sc-ChpC variant was most likely due to unfavorable protein conformations. Moreover, although environmentally sensitive dyes such as TAMRA hardly allowed the comparison of transpeptidation rates via fluorescence intensities, Sm-Chp1 seemed to be a stronger substrate of Sm-SrtE2. Affinity of Sm-SrtE2 to the long chaplins is obviously not only determined by the CWS pentapeptide sequence, but further amino acids in close vicinity to LAXTG. Indeed, Sm-Chp1-like proteins from transglutaminase-producing streptomycetes and the proline-rich repeat-containing protein M3C0P0 (Table 1) harbor a conserved Ala-Gly motif downstream from LAXTG (Fig. S7) that is replaced by Ser/Thr-Asp/Glu in Sc-ChpA-C (Fig. S8).

Details are in the caption following the image
Incorporation of fluorescent Gly3-TAMRA into chaplins by sortase E2. The transpeptidation reaction of 10 μm Sm-Chp1/Sc-ChpC and 50 μm Gly3-TAMRA by 1 μm Δ1–50-Sm-SrtE2 was performed in 50 mm Tris/HCl and 100 mm NaCl pH 7.0 at ambient temperatures. At the indicated times, samples were separated by 12.5% SDS/PAGE, Coomassie-stained or irradiated. (A) Sequence alignment of Sm-Chp1 and Sc-ChpC. (B) Transpeptidation reaction mediated by sortases. (C) Labeling of Sm-Chp1 (left panel) and Sc-ChpC (right panel) during the indicated times. The control (line C) was performed in the absence of sortase.

Interaction of long chaplins and sortase E2 with transglutaminase

Transglutaminase-accessible glutamines of Sm-Chp1 and Sc-ChpC were determined by the enzymatic incorporation of biotin cadaverine (Fig. 5A). The mixtures were separated by SDS/PAGE, blotted and stained by fluorescent streptavidin conjugates (Fig. 5B). In this case, fluorescence corresponds with the number of incorporated biotin molecules and allows semi-quantitative analyses. Using this assay, Sm-Chp1 from S. mobaraensis turned out to be the strongest glutamine donor substrate of Sm-MTG that we ever studied. Biotinylation could only be traced if the E/S ratio was considerably reduced from 1 : 100 to 1 : 5000 (Fig. 5B, left panel). Although biotin cadaverine was used in large quantities, cross-linking of Sm-Chp1 occurred as depicted by the reduced Coomassie-stained protein band and the emergence of a Sm-Chp1 dimer. Sc-ChpC from S. coelicolor was likewise labelled by Sm-MTG but less effectively than Sm-Chp1 (Fig. 5B, right panel). The rapid conjugation with biotin cadaverine led to the assumption that Sm-Chp1 adopts a largely disordered structure in solution, thus allowing MTG unrestricted access to the many glutamine residues (Fig. S8).

Details are in the caption following the image
Biotinylation of long chaplins by transglutaminase. Incubation was performed using 5 μm Sm-Chp1 or Sc-ChpC, 200 μm biotin cadaverine and 0.1 nm Sm-MTG in 50 mm Tris/HCl pH 8.0 at 37 °C. At the indicated times, samples were separated by 12.5% SDS/PAGE, Coomassie-stained or blotted using streptavidin-IRDye-800 conjugates. (A) MTG-mediated transpeptidation reaction and evaluation of incorporated biotin. (B) Biotinylation of Sm-Chp1 (left panel) or Sc-ChpC (right panel) during the indicated times. Control 1 (line C1) shows biotinylated Sm-Chp1 after 60 min, control 2 (line C2) reaction mixtures without Sm-MTG.

We further discovered unexpectedly that Δ1–50-Sm-SrtE2 was biotinylated by Sm-MTG (Fig. 6B, upper panel). Sequence and simulated structure of Sm-SrtE2 suggested that all glutamine positions are inappropriate for the attachment of Sm-MTG (Fig. 1). Although Gln-129 and Gln-176 are putatively integrated in flexible loops, they are surrounded by charged amino acids. Two other glutamines, Gln-62 and Gln-65, were depicted by the Sm-SrtE2 model as part of the N-terminal helix domain, which hardly allows interaction with Sm-MTG. However, if both terminal glutamines were absent, as is the case with the Δ1–78-Sm-SrtE2 variant, biotinylation by Sm-MTG failed (Fig. 6B, second panel).

Details are in the caption following the image
Biotinylation of sortase E2 by transglutaminase. Labeling was performed as described in Fig. 5 at ambient temperature using 5 μm Sm-SrtE2, 200 μm biotin cadaverine and 0.1 μm Sm-MTG. (A) Relative activity of Sm-SrtE2 variants using dabLAETGans as described in Fig. 3. (B) Enzymatic biotinylation of the indicated Sm-SrtE2 variants. (C) Increase in streptavidin-IRDye-800 conjugate fluorescence during the indicated times. Fluorescence intensity was determined by scanning equal protein band areas as shown in (B). The colors for the Sm-SrtE2 variants are consistently used throughout the figure.

To verify this observation, the glutamines were replaced by asparagines via site-directed mutagenesis of Δ1–50-Sm-SrtE2. All prepared variants were active, but Gln-129 in combination with the Gln-178 or Gln-62 seems to have some influence on structural integrity (Fig. 6A). Though the effect of sole Gln-129 substitution on enzyme activity was small in the Δ1–50-Sm-SrtE2-Q129N variant, a strongly enhanced biotinylation rate suggested facilitated access of Sm-MTG to its binding site(s) (Fig. 6B, third panel; Fig. 6C). Further replacement of Gln-62 (Δ1–50-SrtE2-Q62N/Q129N) or Gln-65 (Δ1–50-Sm-SrtE2-Q65N/Q129N) only reduced labeling efficiency, and even in the Δ1–50-Sm-SrtE2-Q129N/Q178N variant enzymatic biotinylation went on, thus clearly showing that Gln-129 and Gln-178 were not the binding sites of Sm-MTG (fourth to sixth panels). Only if both Q62 and Q65 were substituted in the Δ1–50-Sm-SrtE2-Q62N/Q65N/Q129N variant, enzymatic labeling failed (last panel). All these results suggested that both Gln-62 and Gln-65 were equally modified by Sm-MTG, and Gln-129 had some influence on transamidation rate, most likely by reducing flexibility of the N-terminal peptide. Furthermore, it may be concluded from the results that removal of the membrane anchor domain causes disorder in the adjacent peptide, thus providing accessibility of Sm-MTG to Gln-62 and Gln-65.

Discussion

Sortases (Srts) are important enzymes contributing to cell wall and pili assembly by transpeptidation reactions in gram-positive bacteria. Participation of various substrates in pathogenesis has led to the assumption that sortases might be useful targets for controlling microbial infection diseases [11]. Moreover, the transpeptidases have been used for orthogonal conjugation reactions due to their high specificity for distinct pentapeptide motifs, the CWS [28-31].

Type E sortases, first studied in Corynebacterium diphtheriae [32], account for spore development in S. coelicolor as was shown by the deletion of Sc-SrtE1 and Sc-SrtE2 [13]. Both enzymes link the surface-covering long chaplins, Sc-ChpB and Sc-ChpC, to the bacterial cell wall in vivo, and equally cleave in vitro the CWS motifs LAETG and LAHTG [13]. Although less information is available to date on production procedure, purity and stability of the enzymes, crystallization of Sc-SrtE1 has been described as challenging due to irreversible protein precipitation [9].

The present report addresses the biochemical characterization of SrtE2 from S. mobaraensis, which differs from S. coelicolor in many respects but especially in expressing the protein cross-linking enzyme transglutaminase (Sm-MTG) [20, 24]. To this end, three N-terminally shortened Sm-SrtE2 variants and two long chaplins, Sm-Chp1 and Sc-ChpC, were separated from E. coli supernatants with high purity. The most resistant sortase variant, Δ1–50-Sm-SrtE2, readily tended to unfold (melting point of 36–37 °C) and precipitated at moderate temperatures. In general, the more shortened a variant the lower the melting point was (Δ1–501–551–78). Since S. mobaraensis usually exports highly resistant proteins such as Sm-MTG and Sm-MTG substrates tolerating temperatures of 50–80 °C and ethanol up to 80 vol% [33-35], we assume that the removed membrane domain has great influence on structural integrity of Sm-SrtE2. Like Sc-SrtE2, Sm-SrtE2 recognize the CWS motif LAXTG for interacting with its protein substrates. Preference for the predicted binding sites was in the order LAETG>>LAHTG>>LAQTG~LANTG>LARTG. However, the intrinsic substrate Sm-Chp1, accommodating LAHTG, was considerably faster conjugated with Gly3-TAMRA than LAETG-containing Sc-ChpC. The latter protein appears to adopt at least two distinct conformations, one of which obviously deprives Sm-SrtE2 of its access to its binding site, LAETG. The central amino acid of LAXTG influence transpeptidation rate but not the product formation, as was already shown for Sa-SrtA [36].

The usage of substrate peptides such as abzLPXTGdap(dnp) and dabQALPETGEEans has evoked a controversial debate on ‘inner filter effect quenching’ and the correctness of Sa-SrtA kinetic parameters [26, 36-38]. The obtained Km/kcat values amounted to 16–20 μm/0.08–0.44 h−1 or 5.5 mm/972 h−1 by Sa-SrtA-mediated cleavage of the fluorogenic compounds if increase in fluorescence [26, 38] or HPLC peak areas [36, 37] were evaluated. We likewise used the intrinsically quenched peptide fluorophores dabLAXTGans and observed fragment quenching by the complete cleavage of 10–40 μm substrate peptide (Fig. S6). Nevertheless, the amount of 1 μm Sm-SrtE2 sufficiently enhanced fluorescence within the used time span of 10 min. For comparison, Sc-SrtE2 needed overnight incubation (18 h) to determine enzyme activity by cleavage of abzAALAETGSDdnp [13]. Our data, revealing Km of 38–172 μm and kcat of 16–94 h−1 (Table 2), confirmed the high affinity of sortases for the peptidic fluorophores and the low transpeptidation rates. Moreover, enzymatic linkage of Gly3-TAMRA to Sm-Chp1 turned out to be a comparably fast reaction.

Biotinylation of Sm-Chp1 by transglutaminase (Sm-MTG) was still faster and suggested the existence of a completely disordered protein. In the development of aerial hyphae, Sm-Chp1 most likely adopts the functional structure in combination with short chaplins and rodlins during turgor pressure-driven self-assembly of the outer rodlet layer as is described for intrinsically disordered proteins [39]. We assume that ‘folded’ and Sm-SrtE2-anchored Sm-Chp1 still maintains Sm-MTG access to the interconnecting stretches between both chaplin domains as well as the 2nd chaplin domain and LAXTG (Figs S7 and S8). Both intermediate protein regions at least contain a single glutamine that may function like a docking station for the attachment of other Sm-MTG substrate proteins or primary amines. Investigations are in progress to study the interaction of Sm-SrtE2 and Sm-MTG with the long chaplin (Sm-Chp1), a short chaplin (Sm-Chp3), and transglutaminase substrates as building blocks of compressed protein layers.

The biotinylation of Sm-SrtE2 by Sm-MTG was an unexpected result. While failure of Sm-MTG to label Gln-129 and Gln-176 may be the result of backbone interaction and the occurrence of charged and bulky residues in close vicinity [21, 40, 41], the access to Gln-62 and Gln-65 was most likely caused by a largely disordered N-terminal Δ1–50-Sm-SrtE2 peptide (Table S1). Similar flexibility prevented structural analysis of the 30-aa linker segment between the membrane anchor and the catalytic domain of shortened Sc-SrtE1 [9]. We assume that the anchor peptide strongly influences protein stability and should not allow Sm-MTG in vivo to cross-link Sm-SrtE2 with other substrate proteins. However, appreciable effects of Δ1–50-SrtE2 shortening on its biological functions are difficult to estimate. Only truncated sortases have been studied as yet, at least to the best of our knowledge.

Acknowledgements

The study was supported by the German Federal Ministry of Education and Research (grant number: 031A613C). The administration was not involved in study design, in the collection, analysis and interpretation of data, in the writing of the report, and in the decision to submit the article for publication.

    Author contributions

    AA and H-LF conceived and supervised the research; AA and AE designed the experiments; AA, CF, MM and AH-G performed the research; AA, MM, GB-W, HK, and H-LF analyzed the data; AA and H-LF wrote the manuscript with comments from HK.